Expression VTable

[gatesn]

Experiment with new style of vtable.

Overview:

• No more vtable macro.
• "Encoding" struct is now combined with the vtable struct (we had an empty vtable struct before for no reason).
• Instead of Arc, we have an Expression struct with vtable, opaque instance data, and children.
  • We may want this struct to be Arc'd also?
• Because private data can be anything, simple expressions do not need a separate instance struct (e.g. Binary simply stores Operator enum as private data)
• We retain benefits of vtables in that we can run pre/post logic around typed impl functions. We can also expose a smaller API surface on Expression than we do on VTable.

[codspeed-hq]

CodSpeed Performance Report

Merging #5191 will degrade performances by 14.62%

_{Comparing ngates/expr-tree (a5a563e) with develop (45ff246)}

Summary

❌ 4 regressions
✅ 1321 untouched
⏩ 148 skipped¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
❌	pack_return_dtype[1000]	459.2 µs	534.2 µs	-14.04%
❌	pack_return_dtype[100]	50.9 µs	59.6 µs	-14.62%
❌	pack_return_dtype[2000]	916 µs	1,053.6 µs	-13.06%
❌	pack_return_dtype[500]	231.8 µs	269.8 µs	-14.09%

Footnotes

1. 148 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[joseph-isaacs]

can we choose a better name?

[gatesn]

This was here before, agree it's a bit weird but don't really want to change more than I need

[joseph-isaacs]

Is this what we have need calling metadata?

[gatesn]

Kind of...? Metadata sort of implied that it only held onto things that needed to be serialized. Instance data (as in, fat pointer of vtable + data), feels like it's more generic / familiar to the vtable structure.

[joseph-isaacs]

what do you expect to be in data?

[joseph-isaacs]

why add this to the expr, its will be a huge trait if we keep adding

[gatesn]

Anything that needs to be recursively called within a vtable implementation needs to be exposed on the public Expression API.

Agree these names are a bit weird / non-descriptive. Pulling them into a separate trait I don't think fixes that. Maybe stat_max() / stat_min() ?

[joseph-isaacs]

its more about having a trait for analysis stuff that is not on the expr itself

[codecov]

Codecov Report

❌ Patch coverage is 81.70664% with 328 lines in your changes missing coverage. Please review.
✅ Project coverage is 84.96%. Comparing base (b157d2c) to head (a5a563e).
⚠️ Report is 5 commits behind head on develop.

Files with missing lines	Patch %	Lines
vortex-expr/src/exprs/dynamic.rs	1.80%	109 Missing ⚠️
vortex-expr/src/exprs/like.rs	54.76%	38 Missing ⚠️
vortex-expr/src/vtable.rs	81.44%	36 Missing ⚠️
vortex-expr/src/exprs/cast.rs	62.12%	25 Missing ⚠️
vortex-expr/src/expression.rs	86.66%	14 Missing ⚠️
vortex-expr/src/exprs/select.rs	89.24%	10 Missing ⚠️
vortex-expr/src/exprs/list_contains.rs	83.92%	9 Missing ⚠️
vortex-layout/src/layouts/row_idx/expr.rs	50.00%	9 Missing ⚠️
vortex-expr/src/exprs/is_null.rs	79.41%	7 Missing ⚠️
vortex-expr/src/exprs/pack.rs	93.75%	7 Missing ⚠️
... and 22 more

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.